From 966e21ac5542ff39cd6d7abfe9fc2564f5a60b39 Mon Sep 17 00:00:00 2001 From: Youjin Jang Date: Fri, 8 Mar 2024 21:11:22 +0900 Subject: [PATCH] vault backup: 2024-03-08 21:11:22 --- content/NN101/GBO.md | 3 +++ content/NN101/NN101.md | 2 -- 2 files changed, 3 insertions(+), 2 deletions(-) diff --git a/content/NN101/GBO.md b/content/NN101/GBO.md index b9ff7b1..6a335ed 100644 --- a/content/NN101/GBO.md +++ b/content/NN101/GBO.md @@ -53,6 +53,9 @@ We assumed that global minimum is a only local minimum. However, general functio > **Local minimum** is a smallest value in the given range. With method we build in [[GBO#Problem To steep slope|previous chapter]], sequence can easily trapped into the local minimum. +#### Another Observation +![[content/img/GBO-physics.png|250]] +In this example, ball will go to global minimum due to inertia. Therefore, we can apply inertia to our formula. --- [^1]: https://en.wikipedia.org/wiki/Mathematical_optimization diff --git a/content/NN101/NN101.md b/content/NN101/NN101.md index 17733a6..ed93044 100644 --- a/content/NN101/NN101.md +++ b/content/NN101/NN101.md @@ -3,8 +3,6 @@ share: "true" title: NeuralNet101 --- This course explains about basics of AI neural networks. -Main Instructor : [Youjin Jang](https://github.com/jangyoujin0917) - ### Brief Contents - [[GBO|Gradient-Based Optimization]] - Autograd